Rockbox mail archiveSubject: Re: Strong Cryptography
Re: Strong Cryptography
From: Joshua Simmons <mud_at_majidejima.com>
Date: Tue, 20 Nov 2007 19:21:36 -0500
On 11/20/07, Manuel Dejonghe <manuel_at_dejonghe.de> wrote:
> On Nov 21, 2007 12:26 AM, Joshua Simmons <mud_at_majidejima.com> wrote:
> > And finally, at some point I read something about rockbox having a morse
> > code input system, that might be useful to those people who know morse
> That might by the worst idea, since it can be very easy to eavesdrop
> the clicking of buttons. I can prevent someone else watches my hand or
> the screen, but it's hard to make the clicking silent enough.
Yes, you are probably right, that might not be a great idea.
Another problem that comes through my mind is that at least some of
> the algorithms you listed may need division or floating point, which
> may not be available on all platforms. I have myself once tried to
> implement SSL (for a webserver) on a 16 MHz AVR, which I then figured
> out could not work: Single steps of DES took more than seconds, since
> every division had to be replaced by several multiplications etc.
That is certainly a valid concern. I am only interested in implementing
algorithms that can be done quickly and easily in operations that are fast
on the platforms rockbox runs on. As an example, one of the big design
goals for AES (and Twofish as well, since it was an AES finalist) was ease
of implementation of the algorithm in hardware, so no division or floating
point is necessary at all (in fact, large parts of the algorithm can be done
as table lookups). That is not to say that these platforms are going to be
very speedy, but I am very confident that they will be "fast enough" for
relatively small amounts of data (for the password plugin I've mentioned,
we're talking maybe 10 kilobytes, maximum).
Received on 2007-11-21